Thanks for the replies. I wrote a script in SpiderMonkey to place a call, and upon connct turn on vmd, play a "press 1" prompt to see if there is a human, and then play some more prompts. If a voicemail beep is heard it starts playing a different prompt.  I tested this with almost 300 phone numbers, but I suspect a disproportionate number of these are cell phones. Among the calls not answered by a human (nobody pressed 1), about 60% of the calls failed to recognize a beep. My cell phone is T-Mobile, and it doesn't detect the beep for it. 

 

 

 

On 1/8/09, Adam Wilt <wiltingtree@gmail.com> wrote:
> Hi, I have two issues I'd appreciate some help with.
>
> A) I'm testing VMD and I'm getting a success rate of well under 50%. I know
> part of the reason is that some of the voicemail beeps it's encountering are
> very short in length (I've noticed this for T-Mobile and Sprint voicemails,
> and there may be others too), and it can't detect them.  So my question is
> about the notes in the vmd section of the wiki which states, "The industry
> standard is 80% detection. This module if used properly should exceeds the
> standard by a very wide margin".  I'm curious about whether I'm using it
> properly, and what I can do to make it work better.  Thanks for the help.
>
> B) When I place an outbound call and immediately play a prompt when the call
> is answered, the prompt sounds garbled to the person answering the phone. If
> I sleep for a second before playing the prompt, it sounds fine.  Any idea of
> what would cause this?
>
> Thanks,
> Adam
>

Adam,

 Can I ask how you are "testing" vmd?  Where?  How?